The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates shell command templates for running tests using bazel and npm. These commands are configured with environment variables (e.g., SA11Y_AUTO=1) and flags (e.g., --test_output=streamed) necessary for accessibility testing within Salesforce environments. Evidence is found in references/running-sa11y-jest-tests.md which provides templates for single-file, module, and integration testing.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest component names, module names, and file paths from the user prompt to dynamically construct test targets.
Ingestion points: Component and module identifiers provided in the user prompt to populate {moduleName} and {relativePath} placeholders.
Boundary markers: The skill includes environment detection logic (checking for the presence of a WORKSPACE file) to differentiate between core build and standalone execution modes, providing context boundaries for the generated commands.
Capability inventory: The skill uses the npm tool (as declared in the allowed-tools frontmatter) and generates shell strings for bazel test and npm test execution.
Sanitization: Instructions provide strict command templates and specific file-naming conventions (e.g., dropping the .test.js extension) which constrain the input and minimize the risk of arbitrary command injection.

experience-lwc-accessibility-jest-run